Vol.2 No.1 2009

13/88

Research paper : Predictive modeling of everyday behavior from large-scale data (Y. Motomura)−10−Synthesiology - English edition Vol.2 No.1 (2009) References[1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19]S. Russell and P. Norvig: Artificial Intelligence, A Modern Approach, Prentice Hall Series (2002). J. Pearl: Probabilistic inference and expert systems, Morgan Kaufmann,CA, (1988).P. Baldi, P. Frasconi and P.Smyth: Modeling the internet and the web – probabilistic methods and algorithms, (2003). D. Marr: Vision: A computational investigation into the human representation and processing of visual information, W.H.Freeman and Company (1982). G. Cooper and E. Herskovits: A Bayesian method for the induction of probabilistic networks from data. Machine Learning. 9(4), 309-347 (1992).Y. Motomura, T. Iwasaki: Beijian Nettowaku Tekunoroji, Tokyo Denki University. press (2006) (in Japanese). Y. Motomura and T. Kanade: Probabilistic human modeling based on personal construct theory, Journal of Robotics and Mechatronics, 17 (6), 689-696 (2005). Y.Motomura, Y.Nishida: Behavior understanding for everyday support systems in daily environment, Journal of Japanese Society for Artificial Intelligence, 20 (5), 587-594 (2005) (in Japanese). M.Minoh: Human daily life support at a ubiquitous computing home, Journal of Japanese Society for Artificial Intelligence, 20 (5), 579-586 (2005) (in Japanese). B. F. Skinner: Behavior of Organisms, Appleton-Century-Crofts (1938). K.Shiraishi, Y.Yasukawa, Y.Nishida, Y.Motomura, H.Mizoguchi: Information Management Systems for understanding everyday life activity, Annual Conference on Japanese Society for Artificial Intelligence, 3G3-03 (2008) (in Japanese). Y. Nishida, Y. Motomura, G. Kawakami, N. Matsumoto and H. Mizoguchi: Spatio-temporal semantic map for acquiring and retargeting knowledge on everyday life behavior, Lecture Notes in Artificial Intelligence, JSAI 2007 Conference and Workshops, Revised Selected papers, 63-75, Springer-Verlag (2008). G.Kawakami, Y.Nishida, Y.Motomura, H.Mizugochi: Behavior modeling base on spatio-temporal expansion of behavior metrics sensing, Journal of Japan Society for Fuzzy Theory and Intelligent Informatics, 20 (2), 190-200 (2008) (in Japanese). S.Ishikawa, Y.Motomura, S.Kawata, Y.Nishida, K.Hara: Inference and construction of probabilistic causal structure model for everyday life behavior, Annual Conference on Japanese Society for Artificial Intelligence, 3G3-04 (2008) (in Japanese). Y.Motomura, Y.Nishida: Graphical modeling of prior probability in Bayesian estimation for behavior understanding in real life, IPSJ Transactions on Computer Vision and Image Understanding, 18, 43-56 (2007) (in Japanese). Japanese Science and Technology Agency, Center for Research and Development Strategy report, CRDS-FY2005-WR-16(2007) (in Japanese).Y. Motomura: BAYONET: Bayesian network on neural network, Foundation of Real-World Intelligence, 28-37, CSLI calfornia, (2001). Y.Motomura: Bayesian network software BayoNet, SICE Journal of Control, Measurement, and System Integration, 42 (8), 693-694 (2003) (in Japanese). C. Ono, M. Kurokawa, Y. Motomura and H. Asoh: A context-aware movie preference model using a Bayesian adequately respond by transferring the present technology, and the development of additional technology for outcome realization. In the former, the venture responds; in the latter, the choice is made to promote cooperative research between AIST and enterprises.Engineering implementation and societal implementation differ. In engineering, even if the technology is already established, in order to produce societal value, participation of many more stakeholders is necessary. It is necessary to convey value to these stakeholders, which will not necessarily have an engineering background, in order to persuade them to bear the cost and risk; and it is necessary to demonstrate that the outcome has high reliability. Therefore, societal implementation through department-level cooperative research and technology transfer to AIST Ventures is necessary, and results need to be proven in the field. In other words, evaluation of the outcome and societal implementation occurs simultaneously.In order to clarify the conditions under which implementation in society is possible, a marketing research was performed in the Venture Task Force. The cost benefit analysis, which did not need consideration in the first type of basic research, was critical. In order to smoothly advance societal implementation, reductions of cost and risk are sought while improving benefits. At this step, the outcome itself is corrected, and there is a possibility of motivating fundamental research out of necessity for a new outcome. This promotes fundamental research, becomes feedback to fledgling basic research, and is represented in the policy statement of the Digital Human Research Center: “Application driven fundamental research.” It has also become possible to acquire large-scale data that includes situations and context involving the results of activities through actual services and actual users. Bayesian networks constructed from this data forecast the cognitive and evaluation structures and behaviors of existing consumers and others. Being causal models rather than merely descriptive models of the data, they are cognitive models with high reusability and potential for horizontal development (Fig. 7) in other services [23].Concerning issues required for implementation in society, and from the standpoint of fundamental research, whether or not a quick response can be given is thought to be an important issue associated with establishing fundamental research on problem resolutions requested by society in the future. The very fact that speed is requested of technology in society requires that many “buds” be nurtured for the future. Such choices can only be performed by those thoroughly grounded in fundamental research; consequently, in order to perform fundamental research, views aimed at societal technology that clearly envisions the future are surely required.

※このページを正しく表示するにはFlashPlayer9以上が必要です